Dealer banks and the CME squared off in the pages of the Financial Times Sunday, arguing over the systemic risks of CCPs. Each made some good points; each whiffed on a few; both overlooked the real sources of systemic risk arising from clearing mandates.
First, the banks:
"We are losing track of the nominal risk," warned Jean-Pierre Mustier, head of investment banking at UniCredit and a former board member at London-based LCH.Clearnet. "We're moving from a set-up where banks were interconnected, because they had transactions between them, to a system where very lowly capitalised entities, the clearing houses, are supposed to protect the banks from a problem."
. . . .
Banks have voiced concerns to regulators in Europe and the US that central clearing houses are providing insufficient data on their own risks and demanding lower-quality collateral for swaps transactions. They argue the risks are too opaque, and getting riskier.
"It's going in the wrong direction," said the head of credit trading at a large US bank. "The race to the bottom is on." He and other executives were critical of the CME's decision to accept corporate bonds as collateral, a step down from the safer instruments usually required for margin.
There are reasons to be concerned about the expansion of the collateral pool to include more risky assets. This is especially true inasmuch as these assets are likely to become less liquid precisely when a CCP needs to liquidate them: during a systemic event that causes the failure of a large, or several large, clearing members. This problem could be mitigated by giving CCPs access to central bank liquidity, collateralized by the corporate bonds, etc., they take in as initial margin. But the fundamental driver here is that clearing mandates have dramatically increased the need for initial margin, and it was inevitable that this would result in expansion of eligible collateral. It's not a competitive race to the bottom, it's facing reality: indeed, CCPs in individual product spaces face relatively little competition.
Concern about initial margin methodology is also reasonable. Under-margining leads to the mutualization of risk in excess of what CMs had anticipated: over-margining leads to inflated trading costs. But the crucial problem with any CCP methodology is that it takes into account position risk only, and doesn't vary margin with the balance sheet risk of the member firms. This tends to lead to excessive trading by weaker clearing members, as they don't have to post higher collateral, as their poorer credit would suggest they should. This is pretty much an inherent feature of clearing, for informational and governance reasons, as I've discussed frequently in the past.
Terry Duffy, executive chairman of CME Group, said the shift in moving more OTC derivatives transactions was more profound than just transferring the risk and exposure to a counterparty default from banks to clearing houses. "The difference is with a clearing house we do market-to-market [risk management], either twice daily or have the ability to do it much more often," he said on a visit to London.
"What's important is making certain the pays and collects are done on a risk basis and not a mark-to-myth or anything else. Because of that, we have a much easier time doing the pays and collects than the banks do if it was a bilateral transaction. In turn, I think some of the banks are just not used to the model of clearing."
This relates to variation margin, which was / is often collected in OTC transactions (once exposure exceeds negotiated credit limits between the parties.) Several points to make here. First, variation margin is basically a way of keeping initial margin levels current. This doesn't really address the issue raised by the banks of whether initial margin levels are adequate, banks can determine that, and the quality of the collateral posted as initial margin (and whether the haircuts applied against this collateral are commensurate with the risks).
Second, you can go through the process of marking to market, but the protection provided by that process depends crucially on the quality of the prices. That depends on market liquidity, trading activity, etc. You need a market to mark to, and for many products that will be brought into clearing there are serious doubts about the reliability of the prices used for markets. If there's a liquid market, the prices it generates more accurate MTM in both bilateral and cleared trades: if there isn't, both cleared and bilateral MTMs are dodgy. Third, as I'll discuss more in detail, the mechanical process of marking to market can actually create systemic risks.
The FT reporter, Phillip Stafford, also mentions that banks complain about the inadequacy of CCP capital, but notes that CCPs also have default/guarantee funds that they can call on to prevent a CCP default. But that's exactly what the big banks are worried about. Their exposure to CCPs comes first and foremost through their participation in these funds. There's only a thin layer of CCP capital separating a defaulter's margin (and default fund contribution) from the non-defaulters' contributions. If the CCP under margins, that capital can be blown through very quickly, and expose banks to substantial losses. These losses are most likely to occur, moreover, during stressed market conditions when banks are least able to afford the hit. This wrong way risk is precisely what banks are complaining about.
Truth be told, although both the banks and the CCPs (to the extent Duffy is representing the clearing industry perspective) have some valid points, they miss the big systemic risks associated with clearing and clearing mandates. I'm sort of in the middle of my Streetwise Professor Worldwide Personified Clearing Tour, in which I've given talks on the systemic risks of clearing in Chicago, Frankfurt, Osaka, Istanbul, DC, and the UK (London and Oxford), and will give another in Paris in September. (Rough duty, I know, but somebody has to do it.) Here are some of the greatest hits I play at each stop.
- Clearing, and in particular multilateral netting, basically redistributes credit risk from derivatives counterparties to other creditors of defaulting CCP members. These other creditors may be as systemically important and vulnerable, or even more systemically important and vulnerable, than CCP members. For instance, moving to clearing is likely to make runs on money market funds more likely even if it makes runs on dealer banks less likely. It is not evident that this is a desirable trade-off.
- The variation margining mechanism in clearing is particularly rigid, and can create substantial demand for liquidity precisely when funding liquidity dries up. This is what almost brought down the CME clearinghouse on the Day After Black Monday in 1987. This is why rigid marking-to-market and variation margining can be a systemically destabilizing bug, not a stabilizing feature.
- Clearing does not eliminate interconnections between big financial firms: it reconfigures the network of connections. Moreover, it does so in a way that is rife with wrong way risk. As noted before, CCP members-mainly big banks-are connected via CCP default funds, and are most likely to incur losses during periods of market stress, when they are least able to afford it.
Regulators are becoming increasingly aware of these problems, and are responding by attempting to make CCPs virtually immune to failure. But this brings me to another of my greatest hits: beware the fallacy of composition. Yes, failure of a CCP would be catastrophic, but making one part of the financial system stronger- a CCP, for instance- does not necessarily make the system stronger. One of the main effects of increasing the financial resources of a CCP is to redistribute credit losses from derivatives counterparties to other creditors of defaulting firms, and to redistribute scarce liquidity. It is not necessarily the case that this redistribution makes the system stronger. The analogy that I've made here, and that I make on tour, is that of the levee. Building up the levee around a CCP reduces the likelihood that it will be inundated, but forces the financial flood elsewhere. Again, it is not evident that it's better for systemic stability to redistribute the water from CCP members to other parts of the financial system.
My closing number is that CCPs have been vastly oversold as a means of reducing systemic risk. It's virtually impossible to know what the net effects are, but it's likely that CCP mandates reduce systemic risk under some scenarios, but increase it under others. What's more, the main systemic risk is not necessarily the failure of a CCP, as bad as that would be, but the spillover effects from clearing to other parts of the financial system.
That is, it's necessary to analyze the systemic risks of clearing by analyzing the financial system as a whole. Regulators haven't done that, for the most part, and the debate between the banks and Terry Duffy doesn't really do that either. Their's is a CCP-centric debate, and although that's not irrelevant, it doesn't get at the nub of the matte-- meaning that the show must go on.